X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=9207ab76518cfe7c1f09d8cee96b5f249d3771c6;hb=83762ea76e5af65ccb4407c6b38053bc6491a875;hp=92454ed62ef319fdfe69ec81af999c4e9ec66ff0;hpb=cf2735a03e26ba9a2a3aa18c15a720a5ff0bdf72;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 92454ed..9207ab7 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,34 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.17  1995-10-03 14:28:57  adam
+ * Revision 1.26  1995-11-20 11:56:24  adam
+ * Work on new traversal.
+ *
+ * Revision 1.25  1995/11/16  15:34:54  adam
+ * Uses new record management system in both indexer and server.
+ *
+ * Revision 1.24  1995/11/15  19:13:08  adam
+ * Work on record management.
+ *
+ * Revision 1.23  1995/10/27  14:00:10  adam
+ * Implemented detection of database availability.
+ *
+ * Revision 1.22  1995/10/17  18:02:07  adam
+ * New feature: databases. Implemented as prefix to words in dictionary.
+ *
+ * Revision 1.21  1995/10/10  12:24:38  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.20  1995/10/06  13:52:05  adam
+ * Bug fixes. Handler may abort further scanning.
+ *
+ * Revision 1.19  1995/10/04  12:55:16  adam
+ * Bug fix in ranked search. Use=Any keys inserted.
+ *
+ * Revision 1.18  1995/10/04  09:37:08  quinn
+ * Fixed bug.
+ *
+ * Revision 1.17  1995/10/03  14:28:57  adam
  * Buffered read in extract works.
  *
  * Revision 1.16  1995/10/03  14:28:45  adam
@@ -70,12 +97,23 @@
 #include <recctrl.h>
 #include "index.h"
 
+#if RECORD_BASE
+#include "recindex.h"
+#endif
+
 static Dict file_idx;
-static SYSNO sysno_next;
+
+
+#if RECORD_BASE
+static Records records = NULL;
+#else
 static int sys_idx_fd = -1;
+static SYSNO sysno_next;
+#endif
 
 static int key_cmd;
 static int key_sysno;
+static char *key_databaseName;
 static char **key_buf;
 static size_t ptr_top;
 static size_t ptr_i;
@@ -84,8 +122,9 @@ static int key_file_no;
 
 void key_open (int mem)
 {
+#if !RECORD_BASE
     void *file_key;
-
+#endif
     if (mem < 50000)
         mem = 50000;
     key_buf = xmalloc (mem);
@@ -99,6 +138,10 @@ void key_open (int mem)
         logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
         exit (1);
     }
+#if RECORD_BASE
+    assert (!records);
+    records = rec_open (1);
+#else
     file_key = dict_lookup (file_idx, ".");
     if (file_key)
         memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
@@ -109,13 +152,75 @@ void key_open (int mem)
         logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
         exit (1);
     }
+#endif
 }
-    
+
+struct encode_info {
+    int  sysno;
+    int  seqno;
+    char buf[512];
+};
+
+void encode_key_init (struct encode_info *i)
+{
+    i->sysno = 0;
+    i->seqno = 0;
+}
+
+char *encode_key_int (int d, char *bp)
+{
+    if (d <= 63)
+        *bp++ = d;
+    else if (d <= 16383)
+    {
+        *bp++ = 64 + (d>>8);
+        *bp++ = d  & 255;
+    }
+    else if (d <= 4194303)
+    {
+        *bp++ = 128 + (d>>16);
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
+    }
+    else
+    {
+        *bp++ = 192 + (d>>24);
+        *bp++ = (d>>16) & 255;
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
+    }
+    return bp;
+}
+
+void encode_key_write (char *k, struct encode_info *i, FILE *outf)
+{
+    struct it_key key;
+    char *bp = i->buf;
+
+    while ((*bp++ = *k++))
+        ;
+    memcpy (&key, k+1, sizeof(struct it_key));
+    bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
+    if (i->sysno != key.sysno)
+    {
+        i->sysno = key.sysno;
+        i->seqno = 0;
+    }
+    bp = encode_key_int (key.seqno - i->seqno, bp);
+    i->seqno = key.seqno;
+    if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+        exit (1);
+    }
+}
+
 void key_flush (void)
 {
     FILE *outf;
     char out_fname[200];
     char *prevcp, *cp;
+    struct encode_info encode_info;
     
     if (ptr_i <= 0)
         return;
@@ -133,33 +238,19 @@ void key_flush (void)
     logf (LOG_LOG, "writing section %d", key_file_no);
     prevcp = cp = key_buf[ptr_top-ptr_i];
     
-    if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
-    {
-        logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-        exit (1);
-    }
+    encode_key_init (&encode_info);
+    encode_key_write (cp, &encode_info, outf);
     while (--ptr_i > 0)
     {
         cp = key_buf[ptr_top-ptr_i];
         if (strcmp (cp, prevcp))
         {
-            if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
-                        outf) != 1)
-            {
-                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-                exit (1);
-            }
+            encode_key_init (&encode_info);
+            encode_key_write (cp, &encode_info, outf);
             prevcp = cp;
         }
         else
-        {
-            cp = strlen (cp) + cp;
-            if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
-            {
-                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-                exit (1);
-            }
-        }
+            encode_key_write (cp + strlen(cp), &encode_info, outf);
     }
     if (fclose (outf))
     {
@@ -175,8 +266,12 @@ int key_close (void)
 {
     key_flush ();
     xfree (key_buf);
+#if RECORD_BASE
+    rec_close (&records);
+#else
     close (sys_idx_fd);
     dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
+#endif
     dict_close (file_idx);
     return key_file_no;
 }
@@ -198,7 +293,8 @@ static void wordAdd (const RecWord *p)
     ++ptr_i;
     key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
     kused += index_word_prefix ((char*)key_buf + kused,
-                                p->attrSet, p->attrUse);
+                                p->attrSet, p->attrUse,
+                                key_databaseName);
     switch (p->which)
     {
     case Word_String:
@@ -216,6 +312,21 @@ static void wordAdd (const RecWord *p)
     kused += sizeof(key);
 }
 
+static void wordAddAny (const RecWord *p)
+{
+    if (p->attrSet != 1 || p->attrUse != 1016)
+    {
+        RecWord w;
+
+        memcpy (&w, p, sizeof(w));
+        w.attrSet = 1;
+        w.attrUse = 1016;
+        wordAdd (&w);
+    }
+    wordAdd (p);
+}
+
+
 #define FILE_READ_BUF 1
 #if FILE_READ_BUF
 static char *file_buf;
@@ -284,7 +395,8 @@ static int file_read (int fd, char *buf, size_t count)
     return read (fd, buf, count);
 }
 #endif
-void file_extract (int cmd, const char *fname, const char *kname)
+SYSNO file_extract (int cmd, const char *fname, const char *kname,
+                    char *databaseName)
 {
     int i, r;
     char ext[128];
@@ -295,6 +407,7 @@ void file_extract (int cmd, const char *fname, const char *kname)
     struct recExtractCtrl extractCtrl;
     RecType rt;
 
+    key_databaseName = databaseName;
     for (i = strlen(fname); --i >= 0; )
         if (fname[i] == '/')
         {
@@ -308,18 +421,29 @@ void file_extract (int cmd, const char *fname, const char *kname)
         }
     sprintf (ext_res, "fileExtension.%s", ext);
     if (!(file_type = res_get (common_resource, ext_res)))
-        return;
+        return 0;
     if (!(rt = recType_byName (file_type)))
-        return;
+        return 0;
     logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
     file_info = dict_lookup (file_idx, kname);
     if (!file_info)
     {
+#if RECORD_BASE
+        Record rec = rec_new (records);
+
+        sysno = rec->sysno;
+        dict_insert (file_idx, kname, sizeof(sysno), &sysno);
+        rec->info[0] = rec_strdup (file_type);
+        rec->info[1] = rec_strdup (kname);
+        rec_put (records, rec);
+        rec_rm (rec);
+#else
         sysno = sysno_next++;
         dict_insert (file_idx, kname, sizeof(sysno), &sysno);
         lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
         write (sys_idx_fd, file_type, strlen (file_type)+1);
         write (sys_idx_fd, kname, strlen(kname)+1);
+#endif
     }
     else
         memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
@@ -327,22 +451,23 @@ void file_extract (int cmd, const char *fname, const char *kname)
     if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
     {
         logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
-        return;
+        return 0;
     }
     extractCtrl.subType = "";
     extractCtrl.init = wordInit;
-    extractCtrl.add = wordAdd;
+    extractCtrl.add = wordAddAny;
 #if FILE_READ_BUF
     file_read_start (extractCtrl.fd);
 #endif
     extractCtrl.readf = file_read;
-#if FILE_READ_BUF
-    file_read_stop (extractCtrl.fd);
-#endif
     key_sysno = sysno;
     key_cmd = cmd;
     r = (*rt->extract)(&extractCtrl);
+#if FILE_READ_BUF
+    file_read_stop (extractCtrl.fd);
+#endif
     close (extractCtrl.fd);
     if (r)
         logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
+    return sysno;
 }